memorandum to: from: subject10:50 dave hankin (humboldt state university): does random selection and...
TRANSCRIPT
851 S.W. Sixth Avenue, Suite 1100 Steve Crow 503-222-5161 Portland, Oregon 97204-1348 Executive Director 800-452-5161 www.nwcouncil.org Fax: 503-820-2370
Bruce A. Measure Chair
Montana
Joan M. Dukes Vice-Chair
Oregon
Rhonda Whiting Montana
W. Bill Booth
Idaho
James A. Yost Idaho
Bill Bradbury Oregon
Tom Karier Washington
Phil Rockefeller
Washington
MEMORANDUM TO: Chairman Bill Booth and members of the Fish and Wildlife Committee FROM: Tony Grover SUBJECT: Committee discussion of sturgeon management, salmon jacks and sockeye returns The members of the Fish and Wildlife Committee will discuss the following issues of interest to a number of council members: • Discussion of the sturgeon management plan that exists between Washington and Oregon
with an emphasis on harvest. See attached example of the 2010 agreement. • Discussion of the high incidence of Chinook salmon jacks and the pending outcome of a
recent Conference and Workshop on Age and Size at Maturity of Chinook Salmon and other Pacific Salmonids (with emphasis on the Columbia River) held in Portland on May 17-19, 2011. The agenda for the conference is attached. It may be appropriate to consider having the ISAB review the state of the science regarding jack rates.
• Discussion of sockeye returns. A 2008 NOAA report on Sockeye returns to the Columbia
River is attached. Since then, the Fraser River has experience wild swings in numbers of sockeye returning. It may be appropriate to consider having the ISAB review the state of the science regarding sockeye returns.
________________________________________ w:\tg\ww\comm discuss jacks sturgeon sockeye.docx
1
Conference and Workshop on Age and Size at Maturity of Chinook Salmon and other Pacific Salmonids
(with emphasis on the Columbia River)
Date: May 17-19, 2011
Location: Embassy Suites Portland Airport Hotel, 7900 NE 82nd Avenue, Portland, OR 97220
Purpose: Exchange scientific information and empirical data among fishery scientists, managers, and other
interested parties regarding the biological, environmental, and anthropogenic mechanisms affecting age and size
at sexual maturity of Pacific salmon, with special emphasis on hatchery and wild populations of Chinook
salmon in the Columbia River Basin.
Goals of conference and workshop:
Establish a common foundation of knowledge regarding time trends for
age and size at maturity of Chinook salmon in the Columbia River.
Establish a common scientific understanding of the biological,
environmental, and anthropogenic mechanisms that may affect age and
size at maturity of anadromous salmonid fishes.
Identify management issues and problems related to age and size at
maturity of Pacific salmon (e.g., hatchery spawning protocols, run-size
forecasts, etc.).
Identify knowledge gaps and types of data needed to reduce scientific
uncertainties related to management needs and comanager goals.
Identify follow-up tasks to address scientific uncertainties and
management needs.
Intended participants: Fish biologists and managers from government agencies
and tribes, university scientists, and other parties with a scientific or management
interest in the topic.
Scope of Conference and Workshop:
May 17 (12:30-5:40 pm): What do the data tell us? Oral presentations by comanager representatives
summarizing empirical data for hatchery and natural populations.
May 18 (8am-5:40pm): What does the science tell us? Oral presentations by experts summarizing
scientific information on the biological, environmental, and anthropogenic factors that affect age and
size at maturity of Pacific salmon.
May 19 (Invited participation only; 8am-3pm): A working meeting among comanagers and
presenters to address management issues and questions in the Columbia River Basin.
Steering Committee1
Don Campton (Co-Chair), USFWS Kathryn Kostow, ODFW
Doug Olson (Co-Chair), USFWS Joe Krakker, USFWS
Barry Berejikian, NMFS Don Larsen, NMFS
Rich Carmichael, ODFW Brian Leth, IDFG
Ann Gannam, USFWS Steve Schroder, WDFW
Becky Johnson, NPT Brian Zimmerman, CTUIR
1 Affiliations: Confederated Tribes of the Umatilla Indian Reservation (CTUIR); Idaho Department of Fish and Game
(IDFW); National Marine Fisheries Service, NOAA (NMFS); Nez Perce Tribe (NPT); Oregon Department of Fish and
Wildlife (ODFW); U.S. Fish and Wildlife Service (USFWS); Washington Department of Fish and Wildlife (WDFW);
Yakama Nation (YN).
2
Conference and Workshop on
Age and Size at Maturity of Chinook Salmon and other Pacific Salmonids (with emphasis on the Columbia River)
Embassy Suites Hotel at Portland Airport, Oregon
May 17-19, 2011
PROGRAM
DAY 1. Tuesday, May 17, 2011: What do the data tell us?
12:30 pm Welcome – Doug Olson (USFWS)
Session 1. Why are we here?
12:35-1:05 pm Comanager Panel Discussion – Moderator: Joe Krakker (USFWS)
Tim Roth (USFWS): Age and size at maturity – Much ado about nothing or does
it really matter?
Becky Johnson (NPT): Hatchery management questions and issues.
Jay Hesse (NPT): Natural population questions and issues.
Chris Kern (ODFW): Harvest management questions and issues.
Don Campton (USFWS): Science questions and issues.
Questions from audience.
Session 2: Do we have a problem? Is there evidence that the age and/or size at maturity of Chinook salmon (or other species) are changing?
1:10-5:40 pm Case Studies and Populations. Moderator: Rich Carmichael (ODFW)
1:10 Steve Pastor (USFWS) and Doug Olson (USFWS): National Fish Hatchery
observations: variation and trends in age composition and length at return for spring
Chinook, tule fall Chinook, upriver-bright fall Chinook, coho, and steelhead.
1:30 Marc Johnson (ODFW) and Tom Friesen (ODFW): Declines in age and size of
upper Willamette River spring Chinook salmon.
1:50 Andrew Murdoch (WDFW) and Mike Hughes (WDFW): Influence of hatchery
origin spawners on size and age of naturally produced spring Chinook salmon in the
Wenatchee Basin and summer Chinook salmon in the upper Columbia Basin.
2:20 Curt Knudsen (Oncorh Consulting), Bill Bosch (YN), Steve Schroder (WDFW),
Mark Johnston (YN), and Dave Fast (YN): Trends in demographic and phenotypic
traits of upper Yakima River hatchery- and natural-origin spring Chinook salmon.
2:40 Paul Hoffarth (WDFW) and Todd Pearsons (Grant County PUD): Comparison of
size and age at maturity of hatchery and natural origin upriver-bright fall Chinook to
the Hanford Reach.
3:00 Question and Answers: Lower, Mid, and Upper Columbia River panel.
3
3:15 Break
3:30 Debbie Milks (WDFW), Mark Schuck (WDFW), and Bill Arnsberg (NPT): Snake
River fall Chinook - effects of supplementation on population age structure.
3:50 Michael Gallinat (WDFW) and Mark Schuck (WDFW): Tucannon River spring
Chinook - age and size at maturity through 25 years of supplementation.
4:10 Debra Eddy (ODFW), Rich Carmichael (ODFW), Tim Hoffnagle (ODFW), and
Joseph Feldhaus (ODFW): Patterns and trends in age composition and size-at-
maturity for hatchery and natural-origin Imnaha River Chinook salmon.
4:30 Brian Leth (IDFG), and John Cassinelli (IDFG): Age at maturity and length at age
for hatchery and natural populations of spring and summer run Chinook salmon in
Idaho streams.
5:00 Kathryn Kostow (ODFW) and Kevleen Melcher (ODFW): Historic trends in size
and age structure of Chinook salmon captured in Columbia River mainstem fisheries.
5:20 Snake River and Mainstem Fishery Panel: Questions and Answers
5:35 Rich Carmichael (ODFW): Summary and Wrap-up
5:40 pm Adjourn to hotel lounge. Dinner on own.
4
DAY 2: Wednesday, May 18, 2011: What does the science tell us? 8:00 am Introductory remarks – Don Campton (USFWS).
Session 3: What are the effects of hatchery practices on age and size at maturity?
8:05 am Physiology and Nutrition. Moderators: Don Larsen (NMFS) and Ann
Gannam (USFWS).
8:05 Penny Swanson (NMFS): Timing of puberty in salmon.
8:30 Brian Beckman (NMFS): Environmental regulation of puberty in salmon.
8:45 Don Larsen (NMFS): Hatchery influences on puberty in male salmon.
9:00 Pat Connolly (U.S. Geological Survey): Residualization and maturation versus
smolting factors for O. mykiss parr.
9:15 Guillaume Salze (University of Guelph, Ontario, Canada): Dietary modulation of
puberty and the maturation process in fish.
9:30 Ann Gannam (USFWS): Feed changes over time, ingredients and methods of
manufacture, are they having an impact on age/size at maturity of hatchery fish?
9:45 General Discussion
10:00 am Break
10:15 am Genetics and Breeding. Moderator: Don Campton (USFWS).
10:15 Don Campton (USFWS): Heritability of age and size at maturity for salmonid fishes:
Can hatchery spawning practices result in genetic changes over time?
10:35 Christian Smith (USFWS): Precocious males and genetic resources: What if we
remove the jacks?
10:50 Dave Hankin (Humboldt State University): Does random selection and breeding of
adult Chinook salmon in hatchery broodstocks result in earlier age and smaller size at
maturity over multiple generations?
11:20 General Discussion
11:30 am – 12:30 pm: Lunch buffet in hotel lobby (included with registration fee)
5
Session 4: What are the effects of the natural environment, population processes, and anthropogenic factors on age and size at maturity?
12:30 pm Age and size effects on reproductive behavior, breeding success, and
fitness. Moderator: Barry Berejikian (NMFS).
12:30 Steve Schroder (WDFW) and Curt Knudsen (Oncorh Consulting): Effects of size
at maturity on breeding success and lifetime fitness in males and females and
implications for the productivity of natural populations.
12:55 Ewann Berntson (NMFS), Rich Carmichael (ODFW), Robin Waples (NMFS), and
Paul Moran (NMFS): Reproductive success of jacks in natural streams (Catherine
Creek, Wenatchee River).
1:15 Barry Berejikian (NMFS), Steve Schroder (WDFW), and Ewann Berntson
(NMFS): Breeding success of alternative male reproductive phenotypes in Pacific
salmon and evidence for frequency dependence selection.
1:40 General Discussion.
2:00 pm Effects of anthropogenic and environmental factors on age and size at
maturity. Moderator: Kathryn Kostow (ODFW).
2:00 Greg Ruggerone (NR Corporation): Density-dependent growth, maturation, and
survival of Chinook salmon at sea.
2:30 Jeff Hard (NMFS): The demographic and evolutionary implications of harvest for age
and size at maturation in Chinook salmon.
3:00 Break
3:15 Neala Kendall* (UW), Jeff Hard (NMFS), and Tom Quinn (UW): Changes in age
and size at maturity related to fishery selection: lessons learned from Alaskan
populations applied to Columbia River Chinook salmon.
3:35 Tom Cooney (NMFS): Differences in Fishery Selectivity on Male and Female
Columbia River Fall Chinook.
3:55 Steve Haeseker (USFWS): An evaluation of the effects of outmigration experience
on age-at-maturity
4:15 William Connor (USFWS): Variation in age and size at return associated with fall
Chinook salmon juvenile life history.
4:35 Robin Waples (NMFS): The shift toward yearling smolts in Snake River fall Chinook
salmon: evolution or phenotypic plasticity?
4:55 Kathryn Kostow* (ODFW), Henry Yuen (USFWS), and Chris Kern (ODFW): The
quest for environmental and anthropometric variables that explain annual variation in
upriver spring Chinook age structure.
5:15 General Discussion
5:30 pm Adjourn to hotel lounge. Dinner on own. END OF CONFERENCE PORTION OF WORKSHOP
6
NOTE: Day 3 is an invitation-only working meeting among comanagers and presenters to address management issue for hatchery and natural populations in the Columbia River.
DAY 3. Thursday, May 19, 2011: Comanager Workshop. What should we do to achieve management goals for natural and hatchery populations of Chinook salmon in the Columbia River?
Session 5 (3 discussion sessions): What have we done to address management problems related to age/size at maturity, and what should we do in the future?
NOTE: The three sessions below are intended to be flexible to facilitate discussion and debate. 8:00 am Introduction. Tim Roth (USFWS) 8:10 am Session 5a: Hatchery spawning protocols. Moderator: Don Campton (USFWS)
Don Campton (USFWS): Spawning protocol guidelines for National Fish Hatcheries.
Mark Schuck (WDFW) and Bill Young (NPT): Recent changes in brood stock collection
and spawning protocols for Snake River Fall Chinook
Tim Hoffnagle (ODFW), Mike McLean (CTUIR), and Peter Cleary (NPT): Breeding
protocols to minimize effects of age 3 males in spring Chinook supplementation programs
in NE Oregon 9:50-10:05 am Break 10:05 am Session 5b: Hatchery rearing practices. Moderator: Larry Telles (USFWS)
Mark Schuck (WDFW) and Bill Young (NPT): Snake River Fall Chinook size and age at
release
Joseph Feldhaus (ODFW): Size at release studies of Imnaha River spring Chinook to
evaluate how size influences age at return and survivial
Dave Hankin (Humboldt State U.): Effects of month of release (and size at release within month) on maturation schedules and age/size at age.
Don Larsen (NMFS), Curt Knudsen (Oncorh Consulting), Brian Beckman (NMFS),
Bill Bosch (YN), Steve Schroder (WDFW), Mark Johnston (YN), and Dave Fast (YN): Cle Elum Hatchery growth modulation study to reduce high mini-jack rates for Yakima
River Spring Chinook salmon.
Lance Clarke (ODFW): Umatilla Fall Chinook mini-jack study at Umatilla FH:
Production scale evaluation of diet and feeding regimes to influence mini jack rates and
age at return. Presentation would provide background for the problem (too many mini-
jacks) and the experimental design for the study (no data or results yet).
11:45-12:45 Lunch on your own 12:50 pm Session 5c: Escapement management. Moderator: Rod Engle (USFWS)
Tim Hoffnagle (ODFW), Mike McLean (CTUIR), and Peter Cleary (NPT): Control of
natural spawner escapement composition via sliding scale management.
2:30 pm Workshop wrap-up: next steps. Moderator: Don Campton (USFWS 3pm ADJOURN. End of Workshop.
Factors affecting sockeye salmon returns to the Columbia River in
2008
NOAA Fisheries
Northwest Fisheries Science Center
2725 Montlake Blvd. East
Seattle, WA 98112
February 2009
Executive Summary In 2008, more than 213,000 adult sockeye salmon Oncorhynchus nerka returned
to the Columbia River Basin. This is the highest return since 1959. As in the previous 40
years, greater than 99% of these fish were destined for the Upper Columbia River.
Nonetheless, the estimated 805 adults passing Lower Granite Dam marked the highest
return there since 1968.
The high adult sockeye salmon returns in 2008 could have been due to increased
freshwater production, favorable conditions for juvenile sockeye salmon during
downstream migrations, favorable ocean conditions, or a combination of these factors. It
is also possible these high returns resulted in part from reduced harvest or favorable
conditions for adults during their upstream migration to spawning sites. Here we report
analyses of each of these life cycle components to investigate their influence on the high
observed return. This was to analyze the variation in adult return rates across recent
years under contemporary conditions of the mainstem hydropower system. We made no
attempt to relate these returns to those from early periods before or during dam
construction. Direct estimates of smolt-to-adult returns (SAR) for the total Columbia River
population were possible, but for the Snake River population only an “Index SAR” could
be calculated for juvenile outmigrations from 1998 to 2006. Most migrating Snake River
sockeye juveniles were collected at Snake River dams and transported by barge to below
Bonneville Dam, while nearly all Upper Columbia River sockeye juveniles migrated
through the hydropower system and were not transported.
Snake River sockeye Index SARs were substantially lower than Columbia River
sockeye SARs, in part simply because the migration distance incorporated into estimated
SARs was longer for Snake River fish. Snake River Index SARs were estimated from
arrivals of both juveniles and adults at Lower Granite Dam, while the Columbia River
SARs were estimated from arrivals of juveniles at McNary Dam and adults at Bonneville
Dam. A number of additional factors could have influenced the difference in SARs. For
example, adults returning to the Snake River were largely of hatchery origin, while those
from the Columbia River were mostly wild. Furthermore, these stocks are from different
Evolutionarily Significant Units (ESUs). Therefore, we expect them to exhibit inherent
ii
differences that may influence stock productivity. These include genetic differences that
control growth, size, and migration timing. Finally, the Snake River stock represents fish
at the limit of their natural range, with the longest migrations and attaining the highest
elevations to reach spawning sites.
There are several lines of evidence suggesting that changes in ocean productivity
led to the high adult return observed in 2008. Estimated SARs for Upper Columbia and
Snake River sockeye salmon stocks were highly significantly correlated (R2 = 0.87, P <
0.01). In addition, there was no correlation between sockeye salmon SARs and indices of
mainstem flow and percentage spill at hydropower projects in the Columbia River from
McNary to Bonneville Dams. This suggests that the primary factors influencing the
variation in annual adult returns acted downstream from Bonneville Dam and on both
stocks in common. There was no evidence that adult escapements in 2008 were
influenced by changes in ocean or river harvest.
In further support of this finding, we found some evidence that a common
measure of ocean productivity, the Pacific Decadal Oscillation, influenced adult sockeye
returns over a longer time period (1985-2006). Also, the large increase in Columbia
River adult returns from the 2006 juvenile migration, and comparatively large number of
1-ocean fish from the 2007 outmigration, occurred coincident with an increase in ocean
productivity as measured by a suite of indicators developed by the Northwest Fisheries
Science Center for Chinook O. tshawytscha and coho O. kisutch salmon.
Surprisingly, we found a significant negative relationship between survival of
juvenile sockeye salmon from the Upper Columbia River and an index of spill within the
hydropower system between Rock Island and McNary Dams. This finding merits a more
detailed review and analysis of specific project operations and passage conditions.
With respect to the Snake River, the Captive Broodstock Program has increased
smolt production over the past decade, and the number of smolts estimated to have
arrived at Lower Granite Dam correlated strongly (R2 = 0.653, P < 0.01) with the number
of adults returning to Lower Granite Dam two years later. Thus, the large return of adults
to the Snake River in 2008 was in part a result of increased smolt production in 2006.
Also, favorable environmental conditions above Lower Granite Dam in 2008 likely
resulted in a relatively high proportion of adults reaching the spawning grounds that year.
iii
We found no significant correlation between juvenile sockeye survival and indices of
flow, percentage spill, and water temperature. Although an average of 75% of the
juvenile sockeye arriving at Snake River dams from 1998 to 2006 were transported, we
found no significant correlation between Index SARs for Snake River sockeye salmon
and the percentage of fish transported, after adjusting for the SAR pattern common for
both Upper Columbia River and Snake River basin stocks. Unfortunately, we could not
directly compare return rates of PIT-tagged fish with different migration histories -
transported, bypassed, or non-detected - because adult sockeye salmon returns of PIT-
tagged fish were extremely low, and ranged from 0 to 3 fish per treatment group from
juvenile migration years 2002-2007. Given the limited data, it is currently not clear
whether transportation is beneficial, detrimental, or neutral for sockeye salmon.
The analyses conducted here were primarily correlative and limited by the type
and amount of data currently available. Additional research will be required to develop
more robust, definitive information on the factors affecting sockeye salmon in both river
and ocean environments. This would include studies that directly measure the effects of
transportation, evaluate the high variability in smolt survival from traps in the Snake
River to Lower Granite Dam, provide measures of survival past dams and downstream of
Bonneville Dam, and lead to development of ocean productivity indices to predict adult
sockeye return rates.
In summary, the results discussed here provide a consistent pattern to explain the
large return of adult sockeye to the Columbia River in 2008. Based on these results, we
conclude that the factors responsible for the high return largely acted on fish downstream
of Bonneville Dam and during the marine component of their life cycle, and not in the
river upstream of Bonneville Dam.
iv
Introduction In 2008, adult returns of sockeye salmon Oncorhynchus nerka to the Columbia
River were the highest since 1959 (Figure 1). As in each of the past 40 years, greater
than 99% of this return was destined for spawning areas in the Upper Columbia River.
Of the more than 213,000 adults passing Bonneville Dam, only an estimated 805 adults
crossed Lower Granite Dam. However, this return to the Snake River was the highest
observed since 1968.
1940 1950 1960 1970 1980 1990 2000 20100
100
200
300
400
Adult return year
Adu
lt so
ckey
e re
turn
toC
olum
bia
Riv
er (1
000s
)
Figure 1. Yearly adult sockeye salmon return to the Columbia River (Bonneville Dam
count plus Zone 1-5 harvest).
In this report we evaluate factors that likely contributed to the high adult sockeye
salmon returns to the Columbia River in 2008. We examined elements from the entire
life cycle and report on the following areas: freshwater production, conditions during the
juvenile migration, and ocean conditions. We also evaluated the effects on adults of
harvest and conditions experienced during the upstream migration. Because sockeye
salmon that spawn in the Upper Columbia River have constituted nearly the entire return
in recent years and the availability of more long-term data sets, we focused primarily on
this stock. However, because the Snake River sockeye salmon ESU is listed as
endangered under the U.S. Endangered Species Act (NMFS 1991) and operations at
Snake River dams are often scrutinized in terms of their affect on fish survival, we also
evaluated factors affecting this stock. In both rivers, to the extent that data were
available, we evaluated smolt-to-adult return rates (SARs) to measure possible effects of
different factors influencing adult returns.
General sockeye salmon life history
Three Evolutionarily Significant Units (ESU) (Waples 1991) of sockeye have
been identified in the Columbia basin: 1) the Lake Wenatchee ESU; 2) the Okanagon
River ESU that spawns in Osoyoos/Skaha Lakes along the US/Canada border; and 3) the
Snake River ESU (Figure 2). Within these ESUs, three life history types occur:
anadromous, resident and kokanee. The anadromous form spends up to 3 years in its
nursery lake before migrating to sea as a smolt during spring. It may remain at sea up to
4 years before returning to the natal area to spawn (Bjornn et al. 1968; Foerster 1968;
Groot and Margolis 1991). In the Columbia River, most anadromous sockeye spend 1
year in freshwater as juveniles and 2 years at sea as adults. The residual form are
progeny of anadromous or residual fish that remain in fresh water to mature and
reproduce; they produce mostly anadromous offspring (Foerster 1968; Groot and
Margolis 1991; Ricker 1938). Residuals are part of the same ESU as the anadromous
form. From an evolutionarily standpoint, this form may have evolved to act as a safety-
net against failure of other year classes at sea. The third form, kokanee, is a resident,
freshwater-adapted life-history type that evolved from anadromous fish, but is now
genetically distinct from both the residual and anadromous sockeye salmon (Groot and
Margolis 1991).
2
Figure 2. Map of the Columbia River Basin noting the location of major sockeye salmon
spawning lakes (Osoyoos, Wenatchee, and Redfish Lakes) and mainstem dams.
Historically, several populations of Snake River sockeye salmon spawned in
Oregon and Idaho lakes. In the late 19th and early 20th centuries, construction of dams on
headwater reaches of rivers and at lake outlets eliminated access to spawning grounds for
many populations. In Idaho, the anadromous life form was nearly extirpated by
construction in 1910 of Sunbeam Dam on the Salmon River. Currently, Redfish, Pettit,
and Alturas Lakes and the Captive Broodstock Program contain the remnants of the
Snake River sockeye salmon ESU. This population is unique in that it is the
southernmost spawning population in existence. Returning adults from the population
travel the farthest inland (> 1,400 km) and attain the highest elevation spawning grounds
(> 1,980 m) of any sockeye salmon population in the world. Although the population
appeared quite healthy in the 1950s, with hundreds or thousands of fish returning to
3
Redfish Lake most years, it declined precipitously to the point where a total of 3 fish
returned to Redfish Lake during 1988-1990, and the ESU was listed as endangered under
the Endangered Species Act in 1991 (NMFS 1991).
For Redfish Lake sockeye salmon, both the anadromous and residual forms are
beach spawners. These forms reproduce in the lake during October, whereas kokanee
spawn in a tributary to the lake during August and early September. Both the residual
and anadromous forms of sockeye salmon in Redfish Lake are included in the ESA
listing, whereas kokanee in the lake is not. A large population of kokanee also resides in
Dworshak Reservoir, and when spill from the reservoir occurs during spring, large
numbers of juvenile kokanee are often flushed from the reservoir and arrive at Lower
Granite Dam. This complicates estimates of the number of anadromous juvenile sockeye
salmon (smolts) arriving at Lower Granite Dam in a given year.
Adult abundance
For the period 1938-2002, we used estimates of sockeye salmon adult returns to
the Columbia River developed by Washington Department of Fish and Wildlife/Oregon
Department of Fish and Wildlife (WDFW/ODFW 2002). For the period 2003-2008 we
obtained adult counts at Bonneville Dam from University of Washington’s DART web
site (http://www.cbr.washington.edu/dart). We increased these counts by estimated
harvest in the river reach below Bonneville Dam (Columbia River Compact Zones 1-5)
based on unpublished data from Compact reports. Although exhibiting interannual
variability, the number of adults returning to the Columbia River has not displayed any
significant trend over the past 40 years (Figure 1).
Columbia River sockeye
Our analyses of Columbia River sockeye included all adult returns to the
Columbia River and estimates of smolt arrivals at McNary Dam. We recognize that this
mixed the two Upper Columbia River basin ESUs with the Snake River basin ESU, but
because the Snake River ESU produced a very small proportion of these smolts and
4
adults, we assumed that the results essentially reflected factors associated with the Upper
Columbia River stocks. For adult returns, we thus used the data graphed in Figure 1.
Freshwater production The Osoyoos/Skaha Lakes system is more productive and is about 5 times larger
than Lake Wenatchee (Mullan 1986). Consequently, natural smolt production is higher
in Lake Osoyoos than in Lake Wenatchee (12.4 compared to 6.3 lbs fish/acre; Mullan, J.,
pers. comm. as cited in Peven 1987). Smolts leaving Lake Osoyoos are also larger on
average than those leaving Lake Wenatchee (>100 mm and 9.2 g vs. 86 mm and 6.2 g;
Hyatt and Rankin 1999; Peven 1987). Lake Wenatchee fish typically migrate as
juveniles through the mid-Columbia in early May, while Osoyoos/Skaha Lakes fish
migrate in late May or early June (Peven 1987). However, because the Lake Wenatchee
and Okanagon River ESUs share a common migratory pathway and dam counts of smolts
and adults are not distinguished by ESU, we treated all fish from the Upper Columbia as
a single unit in the analyses that follow.
Hatchery enhancement programs have released fry or pre-smolts sockeye into
Lake Wenatchee and Osoyoos/Skaha Lakes the year before they migrate to the ocean
since 1993. The smolts that subsequently migrate from these releases, with exception of
a couple of years (see section below), represent a small fraction of the total sockeye
juvenile migration.
Downstream juvenile migration
We estimated survival of juvenile sockeye from the tailrace of Rock Island Dam
to the tailrace of McNary Dam using standard methods (Skalski 1998; Williams et al.
2001). Release groups were comprised of individual PIT-tagged fish that were tagged or
detected at Rock Island Dam and subsequently detected at or below McNary Dam (i.e.,
John Day Dam, Bonneville Dam, or in the PIT trawl detection system operated at river
kilometer 75). Because of limited sample sizes, survival estimates were not possible for
Upper Columbia River fish in 2003 or prior to 1997. Survival estimates in the remaining
years from 1997 to 2008 ranged from 0.40 to 0.79 (Table 1).
5
Table 1. Survival estimates (with standard errors in parentheses) and environmental
exposure indices for juvenile sockeye salmon migrating from Rock Island Dam to McNary Dam.
Rock Island to McNary
Survival Exposure Indices
S (s.e.)
Flow
(kcfs)
Spill
(%)
Temp
(oC)
1996 -- -- 332.0 32.9 10.7
1997 0.397 (0.119) 478.0 51.8 12.3
1998 0.624 (0.058) 319.8 43.1 12.6
1999 0.559 (0.029) 258.3 46.7 11.1
2000 0.487 (0.114) 242.5 44.3 13.7
2001 0.657 (0.117) 137.0 36.5 15.5
2002 0.531 (0.044) 202.8 35.7 11.6
2003 -- -- -- -- --
2004 0.648 (0.114) 238.1 38.1 14.0
2005 0.720 (0.140) 264.9 40.9 12.7
2006 0.793 (0.062) 321.8 28.7 12.8
2007 0.625 (0.046) 266.0 22.9 13.0
2008 0.644 (0.094) 353.6 26.4 12.3
We evaluated the relationship between juvenile migrant survival and river
conditions within the hydropower system between Rock Island and McNary Dams by
regressing annual estimated survival for the overall Columbia River sockeye salmon
population against indices of the population's exposure to river flow, percentage spill at
dams, and water temperature. Flow and temperature were measured at McNary Dam.
Indices were the average daily flow and temperature between dates of the 25th and 75th
passage percentiles of PIT-tagged fish detected at the dam. For percentage spill we
calculated the average daily percentage spill between the dates of the 25th and 75th
percentiles of PIT-tag detection at McNary Dam, the average daily percentage spill at
Priest Rapids Dam for the period 5 days prior to the McNary passage dates, and the
average daily percentage spill at Wanapum Dam for the period 6 days prior to the
McNary dates. The 5-6 d offset periods were derived from distance between dams and
6
typical travel time for sockeye between Rock Island and McNary Dams. Survival was
significantly negatively correlated with percentage spill, but was not correlated with flow
or water temperature (Figure 3).
100 200 300 400 5000.35
0.45
0.55
0.65
0.75
0.85
R2 = 0.09P = 0.36
McNary Flow Index
Estim
ated
Sur
viva
lR
IS-M
CN
20 30 40 50 600.35
0.45
0.55
0.65
0.75
0.85
R2 = 0.38P = 0.045
Columbia River Spill% Index
Estim
ated
Sur
viva
lR
IS-M
CN
10 11 12 13 14 15 160.35
0.45
0.55
0.65
0.75
0.85
R2 = 0.06P = 0.47
McNary Temperature Index
Estim
ated
Sur
viva
lR
IS-M
CN
Figure 3. Survival of Upper Columbia River sockeye salmon (estimated from Rock
Island Dam to McNary Dam) versus exposure indices for flow (kcfs, top left), spill% (top right), and temperature (o C, bottom left), juvenile outmigration years 1997-2008.
We estimated smolt abundance at McNary Dam - the farthest possible upstream
PIT-tag detection site on the Columbia River. For this estimate, we combined the Smolt
Monitoring Program estimated yearly collection of hatchery and wild smolts at the
McNary Dam juvenile bypass system (JBS) from 1995 to 2007 (http://www.fpc.org). We
divided that total by the estimated annual mean detection efficiency estimates for PIT-
tagged sockeye salmon smolts at the dam. We then expanded this estimate for the recent
years to account for the change in sampling schedule at the JBS from every day to every
other day. We then estimated detection efficiency using CJS (Cormack 1964; Jolly 1965;
7
Seber 1965) maximum-likelihood procedures on data from PIT-tagged sockeye salmon
released from upstream sites in the Columbia River and subsequently detected at
McNary, John Day, or Bonneville Dams. We estimated that since 1995, between 0.43
and 4.76 million smolts arrived at McNary Dam each year (Table 2). The estimated
percentage of wild smolts passing the dam was greater than 95% for all migration years,
except 1996 (81%) and 2005 (89%).
Table 2. Estimated annual number of sockeye salmon smolts from the Upper Columbia
River and Snake River Basins combined arriving at McNary Dam, subsequent adult return (totaled for each outmigration year) to Bonneville Dam, and annual SAR.
Outmigration
Year
Smolts passing
McNary Dama
Adults passing
Bonnevilleb SAR (%)
1995 2,879,662 51,839 1.80 1996 502,153 3,373 0.67 1997 429,543 18,713 4.36 1998 2,856,110 103,178 3.61 1999 3,319,327 112,457 3.39 2000 523,331 49,325 9.43 2001 1,093,809 15,056 1.38 2002 3,118,111 145,741 4.67 2003 4,757,209 73,930 1.55 2004 2,486,746 33,881 1.36 2005 601,769 12,549 2.09 2006 2,849,102 230,665 8.10 2007 2,633,346
a The number of smolts arriving at McNary Dam is comprised of nearly 100% Columbia River stocks due to low production and transportation operations in the Snake River. b Estimated total adults returning to Bonneville Dam from individual juvenile migration years, except for 2006, which was estimated based on returns to date of 1- and 2-ocean fish (203,538) expanded by the long-term proportion of 3-ocean fish (~12%).
There is no information on survival of migrating juvenile sockeye salmon below
Bonneville Dam. However, survival of juvenile Chinook salmon migrating through the
lower Columbia River below Bonneville Dam during spring has been studied since 2005.
Estimated survival for acoustic-tagged yearling Chinook salmon subsampled from the
8
composite population passing Bonneville Dam and pooled across all releases each year
was 0.691 (s.e.= 0.030) in 2005 and 0.671 (s.e.= 0.021) in 2006 (Lynn McComas,
NOAA, NW Fisheries Science Center, pers. communication). While extremely limited,
the available data for yearling Chinook salmon suggest that conditions juvenile sockeye
experienced below Bonneville Dam in 2006 were likely not substantially different from
2005, or responsible for the nearly four-fold difference in estimated adult return rate
between these years (Table 2).
Smolt to adult returns
To estimate smolt-to-adult return rates (SARs) for Columbia River sockeye, we
began with smolt abundance estimates (Table 2) and adult returns to the Columbia River
(Figure 1). Next, we obtained adult age-class composition from Columbia River
Intertribal Fish Commission (CRITFC). Each year, CRITFC staff read scales from a
sample of adult sockeye collected at Bonneville Dam and determine their age at ocean
entry and at adult return to freshwater. Together, the freshwater and ocean ages identify
the age class of a returning adult. For scale analyses from 1998-2008, we accessed the
CRITFC web site (http://www.critfc.org/). J. Fryer (CRITFC, pers. communication)
provided data on adults returning to Bonneville Dam in 1996 and 1997. From these data,
we estimated the proportion of the return each year that came from each juvenile year
class. For each adult return year, we multiplied the estimated proportion of fish in each
age class by the total adults that returned that year (from Figure 1). The resulting adult
return numbers were assigned to the appropriate juvenile year class, and summed to
estimate the total adult return for each juvenile migration year from 1995 to 2006 (Table
2). Not all adults from the 2006 juvenile migration year have returned to date. To
estimate the number of adults returning from this migration year, we expanded the 1- and
2-ocean fish observed to date from 2006 by the mean percentage of 3-ocean fish that
returned to Bonneville Dam over the preceding 10 years (approximately 12%). The
estimated and projected annual SAR of the unmarked Columbia River sockeye salmon
population for migration years 1995 through 2007 ranged from 0.7 to 9.4% (Table 2).
We then evaluated the relationship between Upper Columbia River smolt
abundance and adult abundance from 1995 to 2006 and found a significant (P < 0.05)
9
although relatively weak relationship (R2 = 0.34), suggesting that smolt production did
not explain much of the high return observed in the Columbia River in 2008.
We also evaluated the relationship between SARs and juvenile migration
conditions within the hydropower system downstream from McNary Dam by regressing
annual Columbia River sockeye SAR estimates against two hydropower system
operational indices: 1) an index of flow in the lower Columbia River based on flow at
McNary Dam during the smolt migration season; and 2) an index based on total
percentage spill at McNary, John Day, The Dalles, and Bonneville Dams during the smolt
migration period. The flow index was the weighted average of daily flow values at
McNary Dam, with daily weights equal to the combined number of Snake River and
Upper Columbia River PIT-tagged smolts detected. Weighted average percentage spill
values were calculated for McNary, John Day, and Bonneville Dams using the same
method. The spill index for comparison with SAR was the (unweighted) average of the
weighted averages for the three dams. The annual estimated SAR for the combined
population of fish at McNary Dam had very low correlation and no significant linear
relationship with either flow in the lower Columbia River (R2 = 0.04, P = 0.52) (Figure
4), or the index of spill at McNary, John Day, The Dalles, and Bonneville Dams (R2 =
0.12, P = 0.28) (Figure 5).
100 150 200 250 300 350 400 4500
2
4
6
8
10
points labeled w ithmigration year
2006
2000
1997
2001
19981999
2005
2002
2003
2004
R2 = 0.04P = 0.53
1995
1996
Flow Exposure Index (McNary)
MC
N-B
ON
SA
R (%
)
Figure 4. SAR of the combined Columbia River basin sockeye salmon population
(smolts at McNary Dam to adults at Bonneville Dam) plotted against an index of lower Columbia River flow (kcfs) (based on outflow at McNary Dam) during the juvenile outmigration, 1995-2006.
10
0 10 20 30 40 500
2
4
6
8
10
points labeled w ithmigration year
2006
2000
1997
2001
19981999
2005
2002
2003
2004
R2 = 0.12P = 0.27
1995
1996
Spill% Exposure Index (lower river)
MC
N-B
ON
SA
R (%
)
Figure 5. SAR of the combined Columbia River basin sockeye salmon population
(smolts at McNary Dam to adults at Bonneville Dam) plotted against an index of spill exposure (based on spill percentages at McNary, John Day, The Dalles, and Bonneville Dams) during the juvenile outmigration, 1995-2006.
Relating SARs to ocean conditions Several large-scale ocean and atmospheric indicators influence the coastal ocean
environment, and they in turn affect local physical conditions and the productivity of
coastal waters for juvenile salmon. Large-scale indicators include the Pacific Decadal
Oscillation (PDO) (Mantua et al. 1997) and the Multivariate El Nino Southern Oscillation
(MEI). For the past 12 years, the Northwest Fisheries Science Center has been observing
how these indicators, as well as local and regional physical and biological indicators,
relate to juvenile salmon abundance and subsequent adult returns. We developed an
index of 11 physical and biological ocean ecosystem indicators to monitor the
productivity of the coastal waters, the survival of juvenile salmon, and ultimately adult
returns (http://www.nwfsc.noaa.gov/research/divisions/fed/oeip/a-ecinhome.cfm). To
date, our indices have focused on relating ocean productivity factors with Chinook O.
tshawytscha and coho O. kisutch salmon adult returns.
To address whether SARs for Upper Columbia River sockeye salmon were
related to ocean conditions, we first evaluated whether the available smolt index data at
McNary Dam for migration years 1985-1994 could be used to develop a longer SAR time
series. If SARs based on uncorrected smolt counts (i.e., raw passage index counts) were
11
highly correlated with those based on corrected counts (i.e., passage index counts
corrected with PIT-tag detection probabilities as calculated above), then we could use the
correlation to adjust the early period SARs. In fact, SARs using corrected and
uncorrected smolt counts were highly correlated (R2 = 0.914, P < 0.001), and we
estimated the early period SARs accordingly to expand the time series.
Using the longer SAR time series (1985-2006), we evaluated whether Columbia
River SARs were related to monthly indices of the Pacific Decadal Oscillation (PDO).
The PDO is an index of sea surface temperature anomalies in the North Pacific Ocean
that is commonly used in analysis of how salmon respond to ocean conditions. We
performed a multiple regression of SARs versus monthly PDO indices for each juvenile
migration year. Similar to the approached developed by Zabel et al. (2006) we compared
SARs to monthly PDO indices for a full year beginning in May of the outmigration year.
The combination of months that provided the best fitting model based on Akaike
information criterion (AIC) values contained monthly indices for August and October in
the first year in the ocean and March and April in the second year in the ocean (R2 = 0.61,
P = 0.002). The regression coefficients associated with August (first year) and April
(second year) were negative, meaning cooler ocean temperatures and stronger upwelling
were positively associated with SAR. The coefficients associated with October (first
year) and March (second year) were positive, meaning warmer temperatures and weaker
upwelling were positively associated with SAR. The moderately strong relationship we
obtained indicates that ocean conditions likely play a role in determining SAR, but the
particular months we found important were different from previous analyses (ICTRT and
Zabel, 2007) of Chinook and coho salmon and steelhead, indicating that sockeye are
likely utilizing different ocean productivity factors.
To further explore the recent adult sockeye returns, we compared Columbia River
sockeye SARs to Snake River spring/summer Chinook SARs from 1985 to 2006 to
determine whether common factors were driving the pattern of variability in both spring-
migrating Columbia River basin salmonids. The weak correlation (R2 = 0.16, P = 0.076)
provided additional evidence that Columbia River sockeye were not responding to the
same ocean factors as Snake River spring/summer Chinook salmon. Not surprisingly,
when we compared Columbia River sockeye SARs for migration years 1998-2006 to the
12
suite of factors reported in our ocean index, we found a weak correlation between the
rank order of ocean indicators (a high rank is associated with poor ocean conditions
whereas a low rank is associated with good ocean conditions) and the rank order of
sockeye SARs (R2 = 0.13, P = 0.33). However, we also observed that either higher
sockeye SARs or larger numbers of returning adults were associated with either
intermediate or good ocean conditions, whereas lower SARs or low numbers of returning
adults were associated with poor ocean conditions.
Harvest
Sockeye, pink, and chum salmon rear in the Gulf of Alaska and central North
Pacific Ocean. At one time substantial harvest of all three species occurred in a high seas
driftnet fishery directed at squid. However, based on genetic analysis of fish caught in
this fishery, very few of the sockeye were from southern British Columbia or Washington
(Orlay Johnson, NOAA, NW Fisheries Science Center, pers. communication).
Furthermore, in 1993 the United Nations Convention for the Conservation of
Anadromous Stocks in the North Pacific Ocean prohibited directed fishing for salmonids
in international waters north of 33!N latitude. Today, all regulated harvest occurs within
the Exclusive Economic Zone (EEZ), and most of this is in terminal areas such as
approaches to the Fraser River and Bristol Bay. Thus, we found no evidence that
changes in ocean harvest levels could have resulted in the increased adult return in 2008.
Harvest during the upstream migration, but below Bonneville Dam (management Zones
1-5 under the Columbia River Compact; http://wdfw.wa.gov/fish/crc/crcindex.htm),
made up approximately 0.5% of the total adult return in 2008 and had little influence on
the size of the return.
Upstream adult migration
We found an extremely strong relationship between adult escapement above
Bonneville Dam and adult counts at Rock Island Dam over a broad range of returns over
the past 20 years (Figure 6). Taking into account harvest in the Management Zone 6
fishery between Bonneville and McNary Dams, the average yearly proportion of fish that
successfully migrated between Bonneville and Rock Island Dams since 1978 was
approximately 89%, or roughly 98% survival per dam. The year 2008 was no exception,
where 91% of the fish counted at Bonneville Dam were counted at Rock Island Dam.
13
0 50 100 150 200 2500
50
100
150
200
250
R2 = 0.99P < 0.001
Sockeye salmon escapement (1000s)(Bonneville count minus Zone 6 catch)
Roc
k Is
land
Dam
coun
t (10
00s)
Figure 6. Relationship between upstream escapement (Bonneville Dam count minus the
estimated Zone 6 harvest) and subsequent Rock Island Dam adult count, adult return years 1978-2008.
Snake River sockeye As discussed above, due to the importance of this ESU and its listing under ESA
we evaluated factors that affected adult returns of Snake River sockeye in 2008 and over
the last decade. For these analyses, we used separate smolt and adult abundance
estimates from those used for the Columbia River as a whole. For adult abundance, we
used escapement to upper Snake River dams: Ice Harbor from 1962-1968, Lower
Monumental in 1969, Little Goose from 1970-1974, and Lower Granite from 1975-2008.
These data were obtained from University of Washington’s DART web site
(http://www.cbr.washington.edu/dart). We adjusted counts from 2005 to 2008 when a
removable spillway weir was operated during the summer. Based on PIT-tag detections
in 2008, the weir increased adult counts by approximately 11% due to adults falling back
through the weir and being counted twice. From 1962 to 2008, estimated total adult
returns to the Snake River varied between 1 and 1,300 fish (Figure 7). Patterns of
variation in return numbers were dissimilar from those of Columbia River sockeye as a
whole, and displayed a significant decline until the most recent decade when efforts to
recover the ESU appear to have increased returns compared to the 1985-1998 period
(Figure 1).
14
1960 1970 1980 1990 2000 20100
200
400
600
800
1000
1200
1400
Adult return year
Adu
lt so
ckey
e re
turn
toSn
ake
Riv
er d
am
Figure 7. Yearly adult sockeye salmon counts at upper Snake River Dam (Ice Harbor
Dam (1962-68), Lower Monumental Dam (1969), Little Goose Dam (1970-1974), and Lower Granite Dam (1975-2008). (2006-2008 counts decreased by ca. 11% to adjust for estimated fallback due to operation of the removable spillway weir.)
Freshwater production
A Captive Broodstock Program (Program) was initiated in 1991 to avoid
extinction of Snake River sockeye salmon (Flagg et al. 1995; Flagg et al. 2004). Both
NOAA Fisheries and Idaho Department of Fish and Game (IDFG) maintain Redfish Lake
sockeye salmon captive broodstocks. Groups of fish are reared at two or more facilities
to protect these important genetic lineages from catastrophic loss. IDFG rears captive
broodstock groups until they reach full-term maturity in fresh well water at the Eagle Fish
Hatchery near Boise, ID. NOAA Fisheries rears captive broodstock groups using two
methods, with one group reared to maturity in fresh well water, and a second group
reared from the smolt to adult stages in seawater at its Manchester Research Station near
Port Orchard, WA. All 16 wild anadromous adults that returned to Redfish Lake between
1991 and 1998 were captured and spawned for the Program, and an additional 900 smolts
and 25 residual sockeye salmon were captured for Program rearing in 1991-1993.
15
Today, the Program has first, second, and third generation lineages of these fish in
culture, and efforts are underway to re-introduce populations in nearby Alturus and Pettit
Lakes. The Program reintroduction plan follows a “spread-the-risk” philosophy
incorporating multiple release strategies and sites (Redfish, Alturas, and Pettit Lakes).
Progeny from the Program have been reintroduced to Sawtooth Valley waters at different
life stages using a variety of release methods. Since releases began in the mid 1990s,
more than 860,000 eyed eggs have been planted to in-lake incubator boxes. In addition,
2,500 mature adults were released to spawn naturally, 1.3 million parr were released to
overwinter and migrate the following spring, and 575,000 smolts were released to
migrate immediately. However, releases of different life-history stages during early
years of the Program did not lead to many adult returns. An evaluation in the early 2000s
(Hebdon et al. 2004) determined that returning adults had mostly come from the smolt
releases. Thus, more emphasis in recent years has been placed on smolt production and
release, but production remained low through 2004 due to limited rearing space.
Downstream juvenile migration
We did not use our standard CJS methods to estimate the number of sockeye
smolts arriving at Lower Granite Dam because for several migration years the estimates
based on using CJS methods far exceeded the total number of sockeye smolts estimated
to have left the Stanley Basin. The overestimate of smolts at Lower Granite Dam likely
resulted from kokanee arriving at the dam from Dworshak Reservoir being counted as
sockeye smolts.
Instead, for the 2001-2007 migration years we used estimates of smolts arriving at
Lower Granite Dam made by IDFG and submitted to the Program technical review team
(unpublished data). For 1998-2000, we estimated the number of smolts arriving at the
dam by using estimates of PIT-tagged sockeye salmon leaving Sawtooth Basin traps in
1998 and 1999. We used the average of 1998 and 1999 to represent survival in 2000.
These estimates were then applied to the estimated total number of smolts leaving the
Sawtooth Basin (also made by IDFG and submitted to the Program technical review
team). Since 1998, the estimated number of smolts arriving at Lower Granite Dam has
ranged from approximately 5,300 to 110,000 (Table 3). This wide range reflects the
number of smolts departing Stanley Basin and the wide range in estimates of survival to
16
Lower Granite Dam for smolts caught in traps in the Stanley Basin. From 1994 to 2008,
these estimates ranged from 0.120 (s.e.= 0.013) in 2007 to 0.799 (s.e.= 0.277) in 2005.
We also estimated survival from Lower Granite Dam to McNary Dam for Snake
River juveniles migrating from 1996 to 2008. Because of limited sample sizes survival
estimates were not possible before 1996, or in 1997 and 2004. Estimated survival ranged
from 0.21 (s.e.= 0.063) to 0.76 (s.e.= 0.103) (Table 4). Using PIT-tagged fish released
upstream from Lower Granite Dam, there are several options for estimation of survival
from the tailrace of Lower Granite Dam (LGR) to the tailrace of McNary Dam (MCN).
Particularly for 2006, the choice of estimation method influenced the resulting survival
estimate. A discussion of these options appears in the Appendix. For the analyses that
follow, we used the full data set of all PIT-tagged sockeye released upstream from LGR,
estimating survival from the upstream release point to MCN, then removing the estimate
of survival from release to LGR to obtain the LGR-to-MCN estimate. The Appendix
explains this choice.
Table 3. Production estimates for ESA-listed endangered Redfish Lake sockeye salmon
from Stanley Basin Lakes (combined Redfish, Alturas, Pettit lakes) adjusted to outmigration year for juveniles (all estimates made by IDFG, except as footnoted).
Migration year
Release of pre-smolts
Number of pre-smolts
outmigrating
Number of smolts
released
Release of pre-
spawning adults
Number of eyed
eggs planted
Estimated sockeye juveniles
passing LGR
1998 141,871 61,877 81,615 0 0 96,669a 1999 40,271 38,750 9,718 21 20,311 24,664a 2000 72,114 12,971 148 271 65,200 5,298a 2001 106,166 16,595 13,915 79 0 7,356 2002 140,410 25,716 38,672 190 30,924 16,958 2003 76,788 26,116 0 315 199,666 9,603 2004 130,716 22,244 96 241 49,134 9,749 2005 72,108 61,474 78,330 173 51,239 68,855 2006 107,292 33,401 86,052 464 184,601 109,779 2007 82,105 25,848 101,676 494 51,008 88,398
a Estimated by multiplying survival estimates for Sawtooth trap PIT-tagged sockeye to Lower Granite Dam in 1998 and 1999 (derived using DART) [for 2000, the average of survival estimates from 1998 and 1999] times the IDFG estimate of “Total estimated outmigration”.
17
Table 4. Survival estimates (with standard errors in parentheses) and environmental
exposure indices for juvenile sockeye salmon migrating from Lower Granite Dam to McNary Dam.
Survival Exposure Indices
S (s.e.)
Flow
(kcfs)
Spill
(%)
Temp(o
C)
1996 0.283 (0.184) 158.8 42.5 11.8
1997 -- -- 164.6 39.5 11.4
1998 0.689 (0.157) 130.4 27.4 12.5
1999 0.655 (0.083) 161.2 30.5 11.8
2000 0.679 (0.110) 87.4 28.2 13.8
2001 0.205 (0.063) 61.0 0.0 14.3
2002 0.524 (0.062) 111.7 20.2 11.7
2003 0.669 (0.054) 173.3 37.7 12.3
2004 -- -- 101.3 5.7 12.4
2005 0.388 (0.078) 97.4 8.7 12.7
2006 0.630 (0.083) 168.6 39.5 12.4
2007 0.679 (0.066) 88.1 25.3 12.8
2008 0.763 (0.103) 144.2 40.0 10.6
We evaluated the relationship between juvenile survival and migration conditions
within the hydropower system between Lower Granite and McNary Dams by regressing
annual estimated survival of the Snake River sockeye salmon population against indices
of exposure to river flow, percentage spill at downstream dams, and water temperature.
For each variable, we calculated a average daily value for all dates between the 25th and
75th passage percentiles of PIT-tagged fish detected at each of three dams: Lower
Granite, Little Goose, and Lower Monumental. The Snake River exposure index was
then the average of the three dam-specific indices. No significant correlations (P > 0.05)
were found between survival and any of the conditions evaluated (Figure 8).
18
50 75 100 125 150 175 2000.00.10.20.30.40.50.60.70.80.9
R2 = 0.13P = 0.27
2001
1996
2008
2006
2007
Snake R. Flow Index
Estim
ated
Sur
viva
lLG
R-M
CN
0 10 20 30 40 50 600.00.10.20.30.40.50.60.70.80.9
R2 = 0.30P = 0.08
2001
1996
2008
2006
2007
Snake R. Spill% Index
Estim
ated
Sur
viva
lLG
R-M
CN
10 11 12 13 14 15 160.00.10.20.30.40.50.60.70.80.9
R2 = 0.16P = 0.23
2001
1996
2008
2006
2007
Snake R. Temperature Index
Estim
ated
Sur
viva
lLG
R-M
CN
Figure 8. Survival of juvenile Snake River sockeye salmon (estimated from Lower
Granite Dam to McNary Dam) versus exposure indices for flow (kcfs, top left), spill% (top right), and temperature (oC, bottom left), outmigration years 1996-2008. Data points for key years are noted (high flow-1996; low flow-2001; years the 2008 return outmigrated-2006, 2007).
Smolt to adult returns
For analyses of Snake River SARs, we could not use the same methodology that
we used for Columbia River sockeye, as age-class determinations were not available
from the relatively few adults that returned each year to Lower Granite Dam. Therefore,
we compared all adult returns to Lower Granite Dam with juveniles that migrated from
the dam 2 years earlier for each return year. We assumed ocean-age 2 was the dominant
age class, as it averaged 84% (range 71-97%) of adults sampled at Bonneville between
1987 and 2005. This allowed us to develop a SAR index for migration years 1998 to
2006. This “Index SAR” for the 1998-2006 juvenile migrations ranged from 0.08 to
1.04% (Table 5).
19
Table 5. Estimated outmigration of sockeye salmon smolts from the Sawtooth Basin in
Idaho and adult returns passing Lower Granite Dam.
Out-migration Year
Total estimated sockeye juveniles
Adult return year
Number of adults passing
LGR
Index SAR (%)
1998 96,669 2000 299 0.31 1999 24,664 2001 36 0.15 2000 5,298 2002 55 1.04 2001 7,356 2003 14 0.19 2002 16,958 2004 113 0.67 2003 9,603 2005 18a 0.19 2004 9,749 2006 15a 0.15 2005 68,855 2007 46a 0.07 2006 109,779 2008 805a 0.73 2007 88,398 2009
a adjusted for fallback rate of 11% at Lower Granite Dam estimated from PIT-tag data from adults in 2008. Of 16 PIT-tagged adults passing the dam, 2 passed through the ladders twice, thus 16/18 = 0.89.
We then evaluated the relationship between the Index SARs for Snake River
sockeye salmon and survival estimates for juveniles that migrated from Lower Granite to
McNary Dam and found none (Figure 9). In part, the lack of a relationship likely resulted
from very small sample sizes of juveniles (the starting juvenile population and population
arriving at McNary Dam), as well as small numbers of adults. In addition, the majority
of juveniles were transported and did not experience conditions within the Snake River.
20
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.80.0
0.2
0.4
0.6
0.8
1.0
R2 = 0.15P = 0.35 2006
20032001
2002
2005
1998
2000
1999
points labeled w ithmigration year
Juvenile survival (LGR to MCN)
Inde
x SA
R (%
)
Figure 9. Comparison of annual Snake River sockeye salmon Index SAR estimates with
annual survival estimates of smolts from Lower Granite Dam to McNary Dam, juvenile outmigration years 1998-2006.
We also examined a number of factors possibly related to high returns to the
Snake River in 2008. We correlated annual adult returns from 2000 to 2008 to smolt
abundance in the proceeding two years and found a relatively strong correlation (R2 =
0.66, P < 0.01). Thus, the increased smolt production in 2006 was a likely contributor to
the high return to the Snake River in 2008.
The Index SAR for Snake River sockeye at Lower Granite Dam (Table 6) is a
combined measure for transported and non-transported juvenile fish. During migration
years 1998 to 2006, the estimated percentages of juvenile sockeye arriving at Lower
Granite Dam that were subsequently transported from a dam on the Snake River were 69,
76, 51, 94, 65, 73, 96, 88, and 64%, respectively. To analyze the effects of transportation
on sockeye, we first determined whether data from PIT-tagged sockeye salmon could
provide any reliable information regarding the influence of transportation on adult returns
to the Snake River. For the 2002 through 2007 migration years, fewer than 45,000 PIT-
tagged sockeye salmon were released into the Snake River Basin. Unfortunately, there
was no single year where we could directly compare return rates of PIT-tagged fish with
different migration histories - transported, bypassed, or non-detected - because adult
21
sockeye salmon returns of PIT-tagged fish were extremely low, and ranged from 0 to 3
fish per migration-history group in these years.
Lacking data for direct comparison, we simply correlated the percentage
transported with estimated Index SARs (Table 5) and found a significant negative
correlation (R2 = 0.71, P < 0.01). However, the percentage of Snake River sockeye
salmon transported was also negatively correlated with SARs of Columbia River sockeye
salmon (Table 2) (R2 = 0.73, P < 0.01). Moreover, the three migration years with lowest
percentage of Snake River smolts transported and the three years with greatest Index
SAR were the same (2000, 2002, and 2006), and also coincided with high SARs for the
combined Columbia River population from McNary Dam. In sum, the SAR of sockeye
salmon from the Snake River, which included large proportions of transported fish,
seemed to follow a very similar pattern to the SAR of another group of sockeye salmon
that were not transported at all (Table 2).
The next analysis of transportation effects on Snake River sockeye salmon used
Columbia River SAR as an index of factors the two stocks experienced in common. That
is, we assumed that the effects of common factors were reflected in the Columbia River
SARs. To adjust for common factors, we first regressed Snake River Index SAR on
Columbia River SAR. The linear trend of this relationship was assumed to represent the
effect of common factors. Thus, the residuals from this regression represent an index of
the variation in Snake River SAR that was not explained by common factors, but was
potentially explained by factors unique to the experience of juveniles in the Snake River.
This index of SAR variation had virtually zero correlation with the percentage of fish
transported (R2 = 0.01, P = 0.77) (Figure 10).
22
0.4 0.5 0.6 0.7 0.8 0.9 1.0-0.50
-0.25
0.00
0.25
0.50
2006
2003 20012004
2002
20051999
1998
2000
points labeled w ithmigration year
R2 = 0.01P = 0.77
Proportion transported
Inde
x of
Var
iatio
n of
Snak
e R
iver
SA
R (%
)
Figure 10. Relationship between the proportion of Snake River sockeye salmon
juveniles transported and an index of Snake River-specific variation of subsequent Index SARs (residuals of regression of Snake River Index SARs on Columbia River SARs), juvenile outmigration years 1998-2006.
Relating SARs to ocean conditions and harvest See discussion of Columbia River sockeye, above.
Conditions adults experienced migrating upstream
Conditions experienced by Snake River adults were similar to those experienced by
Columbia River adults up to the mouth of the Snake River, and these conditions were
discussed above in the previous section on Columbia River fish. Since 1984, the annual
number of adults counted at Ice Harbor Dam has not exceeded 100 fish with the
exception of 2000 (216 adults) and 2008 (539 adults). These counts were not
consistently the same, higher, or lower than counts at Lower Granite Dam. Therefore, we
did not attempt to relate adult passage success to environmental conditions in the lower
Snake River. Since 2000, the proportion of fish that have successfully migrated from
Lower Granite Dam to the Sawtooth Basin has varied from 9 to 86% (Table 6). Keefer et
al. (2008) observed that of 31 adult Snake River sockeye salmon radio tagged at Lower
Granite Dam in 2000, all had similar migration rates initially. However, individuals
tagged later in the season eventually slowed their migration and were less successful in
23
reaching spawning areas when water temperatures exceeded 21!C. In 2008, conditions in
the Snake River were generally cooler than in recent years, and the majority of fish
passed Lower Granite Dam when water temperature was " 18!C. Water temperature at
the Anatone gauge on the Snake River near Asotin, WA did not exceed 21!C until 26
July 2008. The majority of fish had likely passed this location by this date, assuming
migration rates were similar to the average 36.8-61.3 km#d-1 reported by Naughton et al.
(2005). Thus, the high variability in successful adult migration since 2000 has likely been
related to environmental conditions, with the high proportion observed in 2008 resulting
from favorable conditions that year. However, these conditions did not have any effect
on the number of adults counted at Columbia or Snake River dams in 2008.
Table 6. Proportion of adult sockeye salmon counted at Lower Granite Dam that successfully migrated to the Sawtooth Basin, 2000-2008.
Year Number of adults passing LGR (data from DART)
Number of adults returning to Sawtooth Basin (data from IDFG) Proportion
2000 299 257 0.86 2001 36 26 0.72 2002 55 22 0.40 2003 14 3 0.21 2004 113 27 0.24 2005 18a 6 0.32 2006 15a 3 0.19 2007 46a 4 0.09 2008 805a 636 0.79 a adjusted for fallback rate of 11% at Lower Granite Dam based on PIT-tag data from adults in 2008. [Of 16 PIT-tagged adults passing the dam, 2 passed through the ladders twice, thus 16/18 = 0.889]
Comparisons between Columbia and Snake stocks Of the analyses conducted, perhaps the most important is the highly significant
correlation between Columbia River sockeye salmon SARs and the Snake River sockeye
Index SARs (R2 = 0.87, P < 0.001) (Figure 11). This provided strong evidence of
common patterns in adult variability, and that factors influencing the variability acted
similarly on both groups of fish. The common factors were evidently not experienced in
24
the juvenile migration phase of the life cycle, as there was essentially no correlation in
estimated juvenile passage survival measured within the hydropower system between the
two groups of fish (R2 = 0.068, P = 0.47). Moreover, there were differences in
correlations of juvenile survival and percentage spill between the two groups. In the
Columbia River, estimated juvenile survival between Rock Island and McNary Dams was
significantly negatively correlated with percentage spill (Figure 3). However, estimated
survival of juvenile Snake River sockeye salmon from Lower Granite to McNary Dams
was positively correlated with percentage spill, but the correlation was not significant
(P = 0.08; Figure 8).
0 2 4 6 8 100.0
0.2
0.4
0.6
0.8
1.0
1.2y = 0.107x - 0.036
R2 = 0.872, P < 0.001
Upper Columbia River SAR (%)
Snak
e R
iver
Inde
x SA
R (%
)
Figure 11. Comparison of estimated SAR for combined Columbia River sockeye salmon
population (smolts at McNary Dam and adults at Bonneville Dam) with Index SAR for Snake River sockeye salmon (smolts and adults at Lower Granite Dam), juvenile outmigration years 1998-2006.
Discussion Adult returns of the composite population of sockeye salmon in the Columbia
River Basin have shown wide fluctuations since 1940. Sockeye SARs also display a high
level of inter-annual variability (Table 2), but fall within the range of SARs for other
sockeye populations from southern British Columbia, Canada, and Lake Washington in
25
Washington State (Koenings et al. 1993). However, in certain years the SARs of
Columbia River Basin sockeye (e.g., 9.4% for migration year 2000; Table 2) were
substantially higher than those of Chinook salmon and steelhead (3.3 and 5% for wild
Snake River spring Chinook salmon and steelhead respectively for migration year 2000;
Williams et al. 2005). This suggests that favorable environmental conditions can result in
a substantial year-to-year increase in sockeye SARs, and the response can differ from
other species. It appears this was the case for juvenile sockeye salmon migrating from
the Columbia River Basin in 2006, when the conditions they experienced that year were
favorable and resulted in an estimated SAR of 8.10 % (Table 2).
The significant correlation between sockeye SARs in both rivers is evidence that
returns in 2008 were most likely influenced by factors downstream of Bonneville Dam.
This result was similar to Peterman et al. (1998), who found that covariation in the
survival characteristics of sockeye salmon was highest amongst stocks that resided in
close proximity to each other. We found no significant evidence that the large returns in
2008 arose from substantial changes in conditions experienced while migrating inriver as
juveniles or adults or from a change in harvest rates in the ocean or river.
Development of ocean ecosystem productivity indicators to date has focused on
the influence of ocean conditions on Chinook and coho salmon returns. The rank order
of indicators developed for other species had a weak correlation (R2 = 0.13) with the rank
order of adult sockeye returns from 1996 to 2008, likely because the indicators did not
incorporate food resources important to juvenile sockeye salmon upon first entering the
ocean. However, on a broader scale, changes in ocean conditions have been shown to
influence Pacific salmon production trends (Mantua et al. 1997; Beamish et al. 1999;
Mueter et al. 2005).
Peterman et al. (1998) also concluded that the early ocean environment affected
the survival of juvenile sockeye salmon and drove observed variation in adult return
rates. Similarly, we observed that for juvenile migration years 1998 to 2006, 4 of 4 years
with either intermediate or good ocean conditions (as predicted by our suite of indicators)
were associated with higher sockeye SARs (greater than 3%) or more than 100,000
returning adult sockeye salmon. In contrast, 4 of 5 years with poor ocean conditions
were associated with SARs lower than 3% or adult returns less than 100,000. We also
26
27
found evidence that ocean productivity as indexed by the PDO influenced adult sockeye
returns from 1985 to 2006.
Snake River Index SARs were much lower than Columbia River SARs, and a
number of factors might have influenced the disparity in return rates. First, the distances
incorporated into the techniques used to estimate the SARs were different between the
two stocks. Snake River Index SARs were calculated from Lower Granite to Lower
Granite Dam, whereas Columbia River SARs were calculated between McNary and
Bonneville Dams. Second, adults returning to the Snake River came largely from the
Captive Broodstock Program while those from the Columbia River were mostly wild.
Wild salmonids typically have higher survival from the smolt to the adult life stage (e.g.,
Williams et al. 2005). Finally, fish from the two areas come from three different ESUs.
Differences in overall SARs between the ESUs were likely affected by differences in
environmental conditions experienced by the ESUs, as well as fundamental genetic
differences, which control growth, size, and migration timing.
Surprisingly, we found a significant negative relationship between survival of
juvenile sockeye salmon from the Upper Columbia River and an index of spill within the
hydropower system between Rock Island and McNary Dams. This finding is in
opposition to a large body of information indicating that increased spill at dams, within
limits, improves the survival of juvenile steelhead and Chinook salmon migrating during
spring (e.g. Muir et al. 2001; Ferguson et al. 2005). It merits a more detailed review and
analysis of specific project operations and passage conditions.
Regarding the transportation of juvenile sockeye from Snake River dams, the low
abundance of juveniles to date has limited the ability to conduct studies that directly
measure the efficacy of transportation. After adjusting for common factors in the patterns
of the Snake River Index SARs and Columbia River SARs, we found no significant
correlation between Snake River sockeye salmon Index SARs and the percentages of fish
transported. Given the limited data, it is currently not clear whether transportation is
beneficial, detrimental, or neutral for sockeye salmon.
The analyses conducted here were primarily correlative and limited by the type
and amount of data currently available. Additional research will be required to develop
more robust, definitive information on the factors affecting sockeye salmon in both river
28
and ocean environments. This would include studies that directly measure the effects of
transportation on sockeye salmon; ascertain whether the high variability in smolt survival
from traps in the Snake River to Lower Granite Dam is the result of natural variability or
other factors; provide measures of survival past dams and downstream of Bonneville
Dam; and lead to development of ocean productivity indices to predict adult sockeye
return rates.
In summary, the results discussed here provide a consistent pattern to explain the
large return of adult sockeye to the Columbia River in 2008. Based on these results, we
conclude that the factors responsible for the high return largely acted on fish downstream
of Bonneville Dam and during the marine component of their life cycle, and not in the
river upstream of Bonneville Dam.
29
References Beamish, R., and coauthors. 1999. The regime concept and natural trends in the
production of Pacific salmon. Canadian Journal of Fisheries and Aquatic Sciences 56(3):516-526.
Bjornn, T., D. Craddock, and D. Corley. 1968. Migration and survival of Redfish Lake, Idaho, sockeye salmon Oncorhynchus nerka. Transactions of the American Fisheries Society 97:360-375.
Cormack, R. 1964. Estimates of survival from the sightings of marked animals. Biometrika 51:429-438.
Ferguson, J., G. Matthews, R. McComas, R. Absolon, D. Brege, M. Gessel, and L. Gilbreath. 2005. Passage of adult and juvenile salmonids through federal Columbia River power system dams. NOAA Technical Memorandum NMFS-NWFSC-64. 160 p. Available from http://www.nwfsc.noaa.gov/publications/.
Flagg, T., C. Mahnken, and K. Johnson. 1995. Captive broodstocks for recovery of Snake River sockeye salmon. American Fisheries Society Symposium 15:81-90.
Flagg, T., and coauthors. 2004. Application of captive broodstocks to preservation of ESA-listed stocks of Pacific Salmon: Redfish Lake sockeye salmon case example. American Fisheries Society Symposium 44:387-400.
Foerster, R. 1968. The sockeye salmon Oncorhynchus nerka. Canada Fisheries Research Board Bulletin 162:1-422.
Groot, C., and L. Margolis, editors,. 1991. Pacific salmon life histories. University of British Columbia Press, Vancouver.
Interior Columbia Technical Review Team (ICTRT) and R. Zabel. 2007. Assessing the impact of environmental conditions and hydropower on population productivity for interior Columbia River stream-type Chinook and steelhead populations. Available at: http://www.nwfsc.noaa.gov/trt/col_docs/matrix_model.pdf
Hebdon, J., P. Kline, D. Taki, and T. Flagg. 2004. Evaluating Reintroduction Strategies for Redfish Lake Sockeye Salmon Captive Broodstock Progeny. American Fisheries Society Symposium 44:401-413.
Hyatt, K., and D. Rankin. 1999. A habitat based evaluation of Okanagan sockeye salmon escapement objectives. Canadian Stock Assessment Secretariat, Research Document 99/19, 59 p.
Jolly, G. 1965. Explicit estimates from capture-recapture data with both death and immigration--stochastic model. Biometrika 52:225-247.
Keefer, M., C. Peery, M. Heinrich, and T. Bjornn. 2008. Behavior and survival of radio-tagged sockeye salmon during adult migration in the Snake and Salmon rivers. Report by U.S. Geological Survey, Idaho Cooperative Fish and Wildlife Research Unit, University of Idaho. (available online at: http://www.cnr.uidaho.edu/uiferl/pdf%20reports/2008-6_SNR%20Sockeye%20Report.pdf.
Koenings, J., H. Geiger, and J. Hasbrouck. 1993. Smolt-to-adult survival patterns of sockeye salmon Oncorhynchus nerka: Effects of smolt length and geographic latitude when entering the sea. Canadian Journal of Fisheries and Aquatic Sciences 50:600-611.
30
Mantua, N., S Hare, Y. Zhang, J. Wallace, and R. Francis. 1997. A Pacific interdecadal climate oscillation with impacts on salmon production. Bulletin of the American Meteorological Society 78(6):1069-1079.
Mueter, F., B. Pyper, and R. Peterman. 2005. Relationships between coastal ocean conditions and survival rates of Northeast Pacific salmon at multiple lags. Transactions of the American Fisheries Society 134:105-119.
Muir, W., S. Smith, J. Williams and B. Sandford. 2001. Survival of juvenile salmonids passing through bypass systems, turbines, and spillways with and without flow deflectors at Snake River dams. North American Journal of Fisheries Management 21:135-146.
Mullan, J. 1986. Determinants of sockeye salmon abundance in the Columbia River, 1880s-1982: a review and synthesis. USFWS Biological Report 86 (12), 137 p.
Naughton, G., and coauthors. 2005. Fallback by adult sockeye salmon at Columbia River dams. North American Journal of Fisheries Management 26:380-390.
NMFS. 1991. Endangered status of Snake River sockeye salmon. Federal Register 56:66(5 April 1991):14055-14066.
Peven, C. 1987. Downstream migration timing of two stocks of sockeye salmon on the mid-Columbia River. Northwest Science 61:186-190.
Peterman, R., B. Pyper, M. Lapointe, M. Adkison, and C. Walters. 1998. Patterns of covariation in survival rates of British Columbian and Alaskan sockeye salmon (Oncorhynchus nerka) stocks. Canadian Journal of Fisheries and Aquatic Sciences 55(11):2503-2517.
Peterson, W.and F. Schwing. 2003. A new climate regime in Northeast Pacific ecosystems. Geophysical Research Letters. 30(17): OCE 6 1-4.
Ricker, W. 1938. “Residual” and kokanee salmon in Cultus Lake. Journal of the Fisheries Research Board of Canada 4:192-217.
Seber, G. 1965. A note on the multiple recapture census. Biometrika 52:249-259. Skalski, J. 1998. Estimating season-wide survival rates of outmigrating salmon smolt in
the Snake River, Washington. Canadian Journal of Fisheries and Aquatic Sciences 55:761-769.
Waples, R. 1991. Definition of "species" under the Endangered Species Act: Application to Pacific salmon. NOAA Technical Memorandum NMFS F/NWC-194, 29 p. (Available online at http://www.nwfsc.noaa.gov).
WDFW/ODFW. 2002. Columbia River Fish Runs and Fisheries 1938-2002. vailable on line at:wdfw.wa.gov/fish/columbia.
Williams, J., S. Smith, and W. Muir. 2001. Survival estimates for downstream migrant yearling juvenile salmonids through the Snake and Columbia rivers hydropower system, 1966-1980 and 1993-1999. North American Journal of Fisheries Management 21(2):310-317.
Williams, J., S. Smith, R. Zabel, W. Muir, M. Scheuerell, B. Sandford, D. Marsh, R. McNatt and S. Achord. 2005. Effects of the federal Columbia River power system on salmonid populations. U.S. Department of Commerce, NOAA Technical Memorandum NMFS-NWFSC-63.
Zabel, R., M. Scheuerell, M. McClure, and J. Williams. 2006. The interplay between climate variability and density dependence in the population viability of Chinook salmon. Conservation Biology 20 (1):190-200.
31
Appendix Notes on Estimation of Survival Between Lower Granite and McNary Dams
For studies using PIT-tagged fish released upstream from Lower Granite Dam,
there are several options for estimation of survival from the tailrace of Lower Granite
Dam (LGR) to the tailrace of McNary Dam (MCN). All options use versions of the
Cormack-Jolly-Seber (CJS) Model of mark-recapture data, but they differ in the
particular set of recapture data used. In this paper’s analyses involving juvenile survival
between LGR and MCN dams, we used the full data set of all PIT-tagged sockeye
released upstream from LGR, estimating survival from the upstream release point to
MCN, then removing the estimate of survival from release to LGR to obtain the LGR-to-
MCN estimate. In this appendix, we discuss the reasons for our choice.
One method that is very familiar from our work with spring migrants in the
survival study program is to select only fish that were detected at LGR and known to
have been returned to the tailrace there. Here we will refer to these fish as “11s,” which
refers to their detection history; the first “1” indicates release upstream from LGR and the
second indicates detection and return to river at LGR. Lower Granite Dam is then used
as the “release site” in a single-release/multiple-recapture (SR/MR) analysis, using
downstream detector dams as recapture sites. The result of this analysis is survival
probability estimates and corresponding standard errors for consecutive river reaches:
LGR to Little Goose Dam (LGO), LGO to Lower Monumental Dam (LMN), and LMN to
McNary Dam (MCN). The product of these estimates is the survival probability estimate
from Lower Granite Dam tailrace to McNary Dam tailrace. The standard error of this
product is calculated from the standard errors and covariances between estimates for
individual reaches. When large numbers of PIT-tagged spring migrants are available, we
group fish detected at Lower Granite Dam into weekly or even daily groups, according to
the date of LGR detection. Each group is then analyzed independently using the SR/MR
model. If fewer PIT-tagged fish are detected and returned to the tailrace at LGR, the date
groupings are sometimes wider, such as biweekly, monthly, or even a single group for the
entire season.
32
A second method uses the same LGR-detected “11” fish as in the method above,
but augments the sample with fish not detected at LGR but detected at Little Goose Dam
(LGO) and returned to the tailrace there. Here we will refer to these fish as “101s,”
referring to their detection history as described above. This strategy is analogous to a
multiple-release/multiple-recapture (MR/MR) design, similar to an avian or mammalian
recapture study where previously untagged animals are trapped, tagged, and added to the
sample on occasions throughout a study. If a date restriction is used, for example weekly
groups of fish in LGR tailrace, this MR/MR approach requires an assumption regarding
travel time from LGR to LGO: if a fish is not detected at LGR and then detected at LGO,
we obviously cannot know for certain which weekly LGR group it belongs to. Only by
assuming a travel time can such a fish be assigned to an LGR weekly group. However, if
only a single group is used for the entire season, then no travel time assumption is
required – all “101s” are combined with all “11s” in the MR/MR analysis. In its summer
2008 memo on sockeye survival, the FPC reported using the MR/MR approach, and
employed a LGR passage date restriction and a travel time assumption for “101s” to build
their sample.
The MR/MR approach can be extended to further “first release sites” at Lower
Monumental Dam (LMN) (“1001” fish released upstream from LGR, first detected at
LMN and then returned to the tailrace) and McNary Dam (MCN) (“10001”s). Further
travel time assumptions are required if a date restriction is applied at LGR.
All told, this results in four SR/MR or MR/MR options that use a “start-at-LGR”
strategy:
(1) Use “11” fish only.
(2) Use “11” and “101” fish.
(3) Use “11”, “101”, and “1001” fish.
(4) Use “11”, “101”, “1001” and “10001” fish.
From a group of tagged fish released upstream at Lower Granite Dam, there are
two detection histories that give no information regarding survival between LGR and
MCN: fish detected and transported from LGR, and fish migrating inriver but never
detected after release. All other detection histories give information. All “start-at-LGR”
33
options use subsets of the entire set of fish that provide information on LGR-to-MCN
survival. Option (1) obviously ignores all fish that are not detected and returned to the
river at LGR. Option (2) adds “101s”, but ignores “1001s” and “10001s”. None of the
options uses fish that are transported after first detection at LGO, LMN, and MCN.
Appendix Table 1 shows the numbers of PIT-tagged fish in various detection
history categories, and the percentages of possible information-giving tagged fish used in
various options using the “start-at-LGR” strategy. On average, only a little over one-
third of tagged sockeye released upstream from LGR were detected and returned to the
river at LGR (i.e., annual average of percentage of “11s” from 1994-2008 was 36.3%).
Moreover, in only one year were there more than 1,000 “11s.” Adding “101s” brings the
annual average up to about 60% of the fish that provide LGR-MCN survival information.
There is a method—call it option (5)—that always uses 100% of the fish that
carry the relevant information. Option (5) is a SR/MR analysis using the group released
upstream from LGR as the single release and estimating survival in four reaches:
Release-to-LGR, LGR-to-LGO, LGO-to-LMN, and LMN-to-MCN. As before, the
product of the estimates (except for release-to-LGR) is the survival probability estimate
from Lower Granite Dam tailrace to McNary Dam tailrace, and the standard error of the
product is calculated from the standard errors and covariances between the estimates for
the individual reaches. This option is attractive when the number of tagged fish is
relatively small, so that the proportion of fish that provide data for the analysis is
maximized. The downside is that it is not possible use these fish and also group the
samples by date of passage at LGR. However, if an annual survival estimate is of most
interest, then the need to maximize sample size may outweigh the inability to restrict by
date.
Given the assumptions of the Cormack-Jolly-Seber model, which underlies both
the SR/MR and MR/MR analyses, all of the methods outlined above give estimates of the
same parameter. That is, expected survival from Lower Granite to McNary Dam is
assumed to be the same for all fish that traverse that section of the river. Thus, the
expected values of estimates arising from all five methods are the same. Of course, there
will be differences in precision of the estimates (i.e., standard errors) related to sample
size, and random variation will result in variations among the estimates given by the
34
different methods. However, there is no a priori reason to suspect that one method will
give estimates that are consistently higher or lower than any other method.
To investigate how much the estimates might vary, we applied each of the five
methods to the annual data from PIT-tagged sockeye released upstream from Lower
Granite Dam. We derived estimates of survival probability between Lower Granite and
McNary Dam using each method (Appendix Table 2). The results showed the expected
trends in precision—using more data usually results in a smaller standard error. With one
or two notably exceptional years, estimates from the various methods were relatively
stable—for any particular year, the range of estimates tended to be similar to the standard
errors of individual estimates. The greatest variability among methods occurred with
2006 and 2008 data. In 2008, all methods except that using only “11” fish gave similar
estimates, but “11” fish amounted to only 18.6% of available data (Appendix Table 1).
In 2006, using only “11” fish (22.3% of available) resulted in an estimate greater
than 1.0, and methods (2) through (5) also resulted in a wide range of estimates. The
observed pattern in the 2006 estimates occurred because fish that were detected for the
first time at McNary Dam were highly significantly more likely detected again
downstream than were fish detected at McNary Dam which had already been detected at
least once at an upstream dam. This violated an assumption of the CJS Model and made
the 2006 estimate problematic. Methods 1, 2, and 3 omit fish detected for the first time at
McNary Dam, and the assumption violation does not occur. However, in all three of
those methods, there were estimates for individual reaches that were greater than 1.0.
Using method (1) LGO-LMN survival was estimated at 1.674 and LGR-MCN survival
was 1.296, as seen in Appendix Table 1. Using method (2) estimated survival was 1.019
for LGO-LMN and 1.036 for LMN-MCN, and using method (3) estimated survival was
1.213 for LMN-MCN. Only methods (4) and (5) give admissible estimates for all
individual reaches, but these methods incur the assumption violation.
For PIT-tagged steelhead and yearling Chinook salmon, we almost always use
option (1) for survival estimates in reaches that begin at Lower Granite Dam. This option
gives maximum flexibility for LGR date restrictions without travel time assumptions, and
the abundant PIT-tag data on these spring migrants sufficiently supports the approach.
The option is clearly less suitable when data are as sparse as for PIT-tagged sockeye.
35
Without a priori reasons to include data from “101” fish but to exclude data from
“1001” and “10001” fish, we see no support for choosing option (2) over option (4).
Given the objective of augmenting data that is otherwise too sparse, we do not see
justification for “going halfway” – if the MR/MR approach with data starting at LGR is
to be used at all, it should use all possible data, including fish detected for the first time at
LMN and MCN. However, we further see no advantage of option (4) over option (5), as
(5) uses the additional data with no additional cost. Thus, in the case of data too sparse
for date restrictions at LGR, our strong general recommendation for estimating LGR-to-
MCN survival is to use option (5).
In the case of the sockeye data analyzed here, the Snake River data for option (5)
(and for option (4)) lead to a violation of an assumption of the CJS model for the 2006
data. In light of this, and on the basis of the reasoning set forth above, we believe that the
decision is not whether to choose between the option (5) estimate of 0.630 (s.e. 0.083)
and the option (2) estimate of 0.850 (0.189), but rather whether to use the option (5)
estimate or to omit the 2006 data point from the analysis entirely. Omission of the 2006
data point makes almost no difference in the nature of the statistical relationships
depicted in Figures 7 and 8 (in particular, R2 values are changed minimally). Therefore,
we opted to leave the 2006 survival estimate of 0.630 in the analyses.
Appendix Table 1. Annual numbers of PIT-tagged sockeye released upstream from Lower Granite Dam, numbers with various detection histories, numbers giving information on Lower Granite-to-McNary survival, and percentage of those providing information used in various analysis options using “start-at-LGR” strategy.
Counts
Percentage of all fish giving information on LGR-MCN survival
Released upstream
Never detected
Trans. LGR
Giving LGR-MCNsurv. info
Det. Hist ”11”
Det. Hist.
“101”
Det. Hist.
“1001”
Det. Hist.
“10001”Det. Hist.
“11” Det. Hist.
“11” + “101”
Det. Hist. “11” + “101” +
“1001” + “10001”2008 6651 5195 8 1448 269 578 369 60 18.6 58.5 88.1 2007 7078 5811 27 1240 471 259 125 152 38.0 58.9 81.2 2006 6431 4690 9 1732 386 720 392 73 22.3 63.9 90.7 2005 8051 5999 501 1551 311 108 30 30 20.1 27.0 30.9 2004 6651 5228 915 508 55 14 12 15 10.8 13.6 18.9 2003 11305 9260 132 1913 707 739 194 102 37.0 75.6 91.1 2002 4603 3629 18 956 255 303 238 81 26.7 58.4 91.7 2001 3296 2556 23 717 603 92 13 3 84.1 96.9 99.2 2000 6359 5107 34 1218 496 277 182 143 40.7 63.5 90.1 1999 8395 6890 177 1328 334 538 283 56 25.2 65.7 91.2 1998 16522 13474 760 2288 1112 458 399 52 48.6 68.6 88.3 1997 1997 1850 4 143 60 26 49 2 42.0 60.1 95.8 1996 8966 7966 143 857 512 150 133 16 59.7 77.2 94.6 1995 4217 4042 36 139 70 21 30 2 50.4 65.5 88.5 1994 767 611 10 146 30 30 21 2 20.5 41.1 56.8
average 1994-2008 36.3 59.6 79.8 average 1998-2008 33.8 59.1 78.3 average 1998-2008, exc. 2004 36.1 63.7 84.3
36
37
Appendix Table 2. Estimated annual probability of survival between Lower Granite Dam and McNary Dam using five different subsets of available data.
Det. Hist. “11”
Det. Hist. “11” + “101”
Det. Hist. “11” +
“101” + “1001”
Det. Hist. “11” + “101” + “1001” +
“10001”
All fish released upstream
Range of survival
estimates (max-min)
Std. Dev. of survival estimates
2008
0.512
(0.138)
0.723 (0.143)
0.761 (0.129)
0.724 (0.101)
0.763 (0.103)
0.251
0.105
2007
0.657
(0.118)
0.613 (0.084)
0.595 (0.074)
0.645 (0.067)
0.679 (0.066)
0.084
0.034
2006
1.296
(0.686)
0.850 (0.189)
0.944 (0.187)
0.668 (0.091)
0.630 (0.083)
0.666
0.267
2005
0.359
(0.104)
0.387 (0.105)
0.423 (0.122)
0.364 (0.077)
0.388 (0.078)
0.064
0.025
2004
NA
NA
0.606 (0.228)
0.757 (0.288)
0.741 (0.254)
0.151
0.083
2003
0.715
(0.114)
0.700 (0.075)
0.701 (0.070)
0.653 (0.055)
0.669 (0.054)
0.062
0.026
2002
0.393
(0.075)
0.428 (0.059)
0.468 (0.063)
0.512 (0.065)
0.524 (0.062)
0.131
0.055
2001
0.248
(0.104)
0.205 (0.068)
0.212 (0.071)
0.200 (0.062)
0.205 (0.063)
0.048
0.019
2000
0.560
(0.142)
0.594 (0.126)
0.584 (0.113)
0.645 (0.114)
0.679 (0.110)
0.119
0.048
1999
NA
0.518
(0.090)
0.606 (0.092)
0.629 (0.083)
0.655 (0.083)
0.137
0.059
1998
NA
0.847
(0.345)
0.721 (0.227)
0.674 (0.157)
0.689 (0.157)
0.173
0.079
Average Std. Err.
0.185
0.128
0.125
0.105
0.101